Unsupervised Representation Learning With Long-Term Dynamics for Skeleton Based Action Recognition

Authors: Nenggan Zheng, Jun Wen, Risheng Liu, Liangqu Long, Jianhua Dai, Zhefeng Gong

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We quantitatively evaluate the effectiveness of our learning approach on three well-established action recognition datasets. Experimental results show that our learned representation is discriminative for classifying actions and can substantially reduce the sequence inpainting errors.
Researcher Affiliation Academia Nenggan Zheng,1 Jun Wen,2 Risheng Liu,3 Liangqu Long,2 Jianhua Dai,4 Zhefeng Gong5 1 Qiushi Academy for Advanced Studies, Zhejiang University, Hangzhou, Zhejiang, China 2 College of Computer Science and Techology, Zhejiang University, Hangzhou, Zhejiang, China 3 DUT-RU International School of Information Science & Engineering, Dalian University of Technology, Liaoning, China 4 College of Information Science and Engineering, Hunan Normal University, Changsha, Hunan, China 5 Department of Neurobiology, Zhejiang University School of Medicine, Hangzhou, Zhejiang, China
Pseudocode Yes Algorithm 1 Training the conditional inpainting model.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository for the methodology described.
Open Datasets Yes We perform our experiments on the following three datasets: the CMU dataset (CMU 2003), the HDM05 dataset (M uller et al. 2007), and the Berkeley MHAD dataset (Ofli et al. 2013).
Dataset Splits Yes For the entire dataset, the testing protocol is 4-fold cross validation, and for the subset, it is evaluated with 3-fold cross validation. [...] We follow the experimental protocol proposed in (Du, Wang, and Wang 2015) and perform 10-fold cross validation on this dataset.
Hardware Specification No The paper does not specify any hardware details such as CPU/GPU models, memory, or cloud instances used for running the experiments.
Software Dependencies No We implement our model in Tensorflow (Abadi et al. 2016) and optimize it with ADAM (Kingma and Ba 2014). (No version numbers provided for TensorFlow or ADAM).
Experiment Setup Yes We set dropout ratio to be 0.2. [...] We find that smaller λadv helps the Enc to learn a more effective representation, and we set it 0.1 in experiments. [...] Parameters λz controls the weight of z in the total adversarial loss... we set it 0.1. [...] Each layer of the Enc and Dec has 800 hidden units. The Dis network is smaller, with 200 hidden units each layer.